JP4751886B2 - Image judgment method - Google Patents

Description

The present invention relates to an image determination method for determining the appearance of a predetermined subject in video and images taken by a camera, and particularly to a case where another building exists between the camera and the subject. However, the present invention relates to an image determination method capable of selecting optimal video and image data in consideration of the other building.

  Conventionally, when a user who uses video data concerning a predetermined subject (hereinafter simply referred to as a user) acquires video data and images of the subject, a plurality of cameras installed at different positions are used. The subject can be photographed from various directions, and the user himself / herself browses and selects the most suitable data from the photographed video and image group, that is, the data that best reflects the subject, and uses the selected video and image data. Was.

  However, in the above-described method, when the number of video images and image data of subjects captured by a plurality of cameras increases, the user checks each video and image data one by one to check the image frame in the optimal video. In addition, since the user himself / herself needs to select image data (hereinafter simply referred to as an image), there is a problem that the burden on the user is large.

  Therefore, in order to reduce the burden on the user, the optimal video and image data can be automatically selected based on the camera's shooting space and subject's positional relationship using the camera's shooting position and direction and subject position. (See, for example, Patent Document 1 and Patent Document 2). According to these techniques, as shown in FIG. 22, with respect to an image captured by the camera A, an elliptical cone having a position at the camera A as a vertex and spreading at a camera angle of view calculated from a focal length or the like with the imaging direction as the center. Alternatively, a square pyramid space is defined as an imaging space, and it is determined that the subject AB in the imaging space is reflected in the corresponding image, and that the subject C outside the space is not reflected.

  Further, as a method for determining the degree of projection of the video and the image, the distance from the camera A is obtained for each subject AB determined to be reflected in the image of the camera A, and the subject B with a short distance is reflected. It is determined that the subject is good, and it is determined that the subject A with a long distance is poorly reflected.

  In addition, as a method for determining the degree of reflection of other images and images, the state of reflection of the subject A existing near the central axis of the elliptical pyramid or the quadrangular pyramid in the shooting space, among the subjects AB existing in the shooting space of the camera. There is also a method of determining that the quality is good.

JP 2001-34250 A JP 2003-179908 A

  However, such a conventional technique has a problem that it is not always possible to select optimal video and image data.

  This is because, as shown in FIG. 23, for example, there is another building object group between the camera A and the subject AB, the subject AB is hidden behind these other building object groups, and the camera A appropriately controls the subject AB. This is because even if the image cannot be captured, the conventional method erroneously selects the image captured by the camera A as the optimal image data for the subject AB.

  This erroneous selection occurs even in a method of selecting an image by determining the appearance of the subject, and the conventional technique does not take into account other buildings existing between the camera and the subject. Image data cannot be selected properly.

  In other words, even when another building exists between the camera and the subject, it is extremely important to be able to automatically select the optimum image data in consideration of the other building. Yes.

The present invention has been made in order to solve the above-described problems caused by the prior art, and even when other buildings exist between the camera and the subject, the other buildings are taken into consideration. An object of the present invention is to provide an image determination method capable of automatically selecting optimal image data.

In order to solve the above-described problems and achieve the object, the present invention includes a step in which a computer extracts a first feature point from among a plurality of feature points related to the shape of a subject based on information relating to a region where the subject exists. , On the first line segment based on the first line segment connecting the position of the first feature point and the viewpoint position of the camera that captures the subject, and information on the existence area of the other subject. Based on the step of determining whether or not another subject exists and the second line segment for the second feature point different from the first feature point, the other subject is the second line segment. A step of determining whether or not the image is present above and a step of calculating a ratio at which the photographing of the subject by the camera is shielded by the other subject based on the result of the determination. To do.

According to the present invention, the computer extracts a first feature point from among a plurality of feature points related to the shape of the subject based on information relating to the region where the subject exists, the position of the first feature point, and the Whether or not the other subject exists on the first line segment is determined based on the first line segment that connects the viewpoint positions of the camera that shoots the subject and information on the existence area of the other subject. And determining whether or not the other subject exists on the second line segment based on a second line segment for a second feature point different from the first feature point And a step of calculating a rate at which the photographing of the subject by the camera is shielded by the other subject based on the determined result, and how much the subject and the obstacle overlap. Can be determined. Also, it is possible to select optimal image data with a good image quality of a predetermined subject.

Examples of the image determination method according to the present invention will be described below.

  First, the principle of the image determination method in the present invention will be described. FIG. 1 is a principle diagram showing the principle of an image determination method according to the present invention. The figure shows a configuration of an image determination method 100 that calculates, as an other object occlusion index, how much other object occlusion is received by a subject (hereinafter referred to as an object to be determined) in a video and an image. At the same time, the data used by the image determination method 100 is shown.

  The image determination method 100 includes a subject shape estimation unit 120 that mainly performs processing related to a determination target subject, and a subject feature point extraction unit as pre-processing for starting the video and image specific image occlusion determination by blocking other objects. 130, and includes an other object occlusion determination unit 140 and an other object occlusion index calculation unit 150 that process the image quality determination itself in each image frame of a video or an image using the result of the preprocessing. Usage data includes subject area position information 110d, which is data relating to a determination target subject to which the image determination method 100 is applied, camera data 110e relating to a camera that has captured a video and an image, and preprocessing and main processing of this method. There is building / surface height map information 110 to be performed.

  The subject area position information 110d is position information of the existence area of the determination target subject. The existence area can have various designations such as a single point, an ellipse, a rectangle, and a polygon. For example, when the existence area is a polygon, position coordinates such as the latitude and longitude of each vertex constituting the shape. It consists of value, structure information (topology) necessary for constructing a polygon from each vertex. The existence area is used to specify an occupied area on each map of the determination target subject.

  The building / surface height map information 110b includes various height information such as the height of a natural object (hereinafter referred to as a building) such as an artificial structure or a tree, and the height of a pure surface excluding the structure. The information is a map together with the measurement position. Specific examples of holding the map information will be described in detail in the first and subsequent embodiments. However, these various heights may be held separately or the added heights may be held. Here, the map is expressed for the purpose of understanding that various heights can be acquired from the position information. However, when actually mounted on the apparatus, it can be stored as a table list. FIG. 2 is a diagram illustrating an example of the building / surface height map information 110b.

  The camera data 110e is information of a shooting camera related to each image frame of a video and an image to which the present invention is applied, camera parameters such as a camera position, a shooting direction, a focal length, and an angle of view that can be calculated from them. Information. In the present invention, each image frame of the video and the image is not directly used for the determination, but the camera data 110e related to them is used for the image determination, and the video data corresponding to the camera data 110e having a good determination result is finally obtained. Are selected as the optimum video data related to the subject to be judged.

  The subject shape estimation unit 120 estimates the subject shape by combining the subject region position information 110d and the building / surface height map information 110b to obtain the height of the region where the determination target subject exists. The method of extracting the approximate shape of the determination target subject is not limited to the above-described method, and the approximate shape of the determination target subject is determined using detailed three-dimensional data such as CAD (computer aided design) data. It may be extracted.

  The subject feature point extraction unit 130 calculates the position and height of the feature point representing the determination target subject based on the shape estimated by the subject shape estimation unit 120. The calculation method is not particularly limited as long as it is a feature point representing the position and shape of the determination target subject. For example, as described in the first embodiment, the shape may be divided into meshes with an arbitrary width and mesh grid points may be used, feature points may be randomly acquired from within the shape, or one center of gravity of the shape May be used as a feature point.

  The other object occlusion determination unit 140 performs the building / surface height for each point on a line segment connecting the feature point extracted by the subject feature point extraction unit 130 and the camera position of the determination target that is the camera data 110e. By acquiring the height information in the map information 110b and comparing the height with the height of the line segment, the presence of the subject object exceeding the line segment is detected. Further, it is determined whether or not the subject object is the determination target subject by determining whether or not there is a point position where the subject object exists in the determination target subject existing area of the subject region position information 110d. It is determined whether or not. If there is no subject object exceeding the line segment at all points on the investigated line segment, it is determined that the corresponding feature point is not obstructed by anything. The other object occlusion determination unit 140 classifies the feature points in these determinations into three types: occluded by other objects, occluded by the determination target subject itself, and not obstructed by anything.

  The other object occlusion index calculation unit 150 uses the determination classification result for each feature point determined by the other object occlusion determination unit 140, and the image condition related to the other object occlusion in the video and image related to the corresponding camera. An index value representing is finally calculated.

  As described above, the image determination method 100 extracts the feature points of the determination target subject, and how much the feature points viewed from the camera are hidden by other subjects based on the extracted feature points and the heights of the other subjects. And the user determines the appearance of the subject to be determined in the image, and based on the determination result, the user selects the optimum video and image data from a plurality of videos and image data captured by a plurality of cameras. Can be selected efficiently.

  Next, as a specific example of the principle diagram of the image determination method in the present invention, Example 1, which is a configuration example of a device (hereinafter referred to as an image determination device) equipped with an image determination method, will be described in detail with reference to the drawings. explain. FIG. 3 is a functional block diagram illustrating the configuration of the image determination apparatus according to the first embodiment. As shown in the figure, in addition to the principle diagram of the present invention shown in FIG. 1, the image determination device 10 includes an input device 40 and a display device 50, vector map information 110a, a conventional subject presence determination unit 60, video data. 110 c and a display processing unit 160. In addition, as an example of the building / ground surface height map information 110b, the building / land surface height map information 100ba and the ground surface height map information 110bb are included.

  The input device 40 is an input device such as a keyboard and a mouse, and the output device 50 is a display device such as a display.

  The vector map information 110a is a general vector map as shown in FIG. 4, and indicates an area on the map where a subject such as a building exists. In the first embodiment, an example is shown in which the user designates the position of the determination target subject area using this vector map, and uses the vector graphic that matches the designated area position as it is as the subject area position information 110d.

  FIG. 5 is a diagram illustrating an example of display using a vector map that is displayed on the display device 50 when the apparatus equipped with the image determination method of the first embodiment starts using the image determination method. Based on the vector map displayed on the display device 50, the user designates a determination target subject, selects a vector shape on the vector map, and position information such as latitude and longitude of each point of the shape, that is, determination The contour point of the target subject area is regarded as the subject area position information 110d. At this time, the user uses the input device 40 to trace and specify the determination target subject existing on the vector map displayed on the display device 50, and selects the vector shape at that position.

  Note that the method of acquiring the subject region position information 110d using the vector map information 110a is not limited to the method shown in FIG. 5, and for example, as shown in FIG. The neighboring vector shape may be designated by, or as shown in FIG. 7, the neighboring vector shape may be designated by enclosing the determination target subject with a polygon. Alternatively, names, addresses, telephone numbers, and the like that are associated with vector graphics on a vector map in advance may be prepared, and vector graphics may be designated by names, addresses, telephone numbers, and the like.

  Of course, it is also possible to acquire the object region position information 110d without using the vector map information 110a as in the first embodiment, and the contour of the subject region to be determined is directly designated by the latitude and longitude numerical values. Or, instead of the vector map of FIG. 4, a picture taken in the surrounding area or a picture drawn is displayed, and a drawing position in the input device is drawn while drawing an outline with the input device similar to FIG. May be specified by directly converting to latitude and longitude.

  In the first embodiment, the building / surface height map information 110b is, for example, a digital map of the height of the ground surface including the building, which is a digital map of the height of the ground surface including the building, and the height of the pure ground. It is composed of the height map information 110bb of the ground surface, which is a digital map. The building litter surface height map information 110ba is information that holds the height of the building when the building is located, and the height of the pure ground when there is no building, together with the location. This configuration and content is an example, and each element is divided and held, such as a height map of only artifacts in a building, a height map of only trees in a building, a height map of only the ground surface, etc. These may be substituted, and these may be divided and held in administrative division units or the like.

  The camera data 110e is a data group in which camera data such as the position and direction of the camera and the angle of view when a video and an image are captured by a plurality of cameras (not shown).

  The video data 110c is video and images taken by a plurality of cameras (not shown). The video data 110c and the camera data 110e are held so as to have a mutual relationship. The image determination apparatus 10 according to the first embodiment does not directly determine the video data 110c, but includes the image determination method 100 using the related camera data 110e, so that the determination specified by the user from the input device 40 is performed. The object is to select the optimum video data relating to the target subject from the video data 110c.

  The conventional subject presence determination unit 60 is a processing unit that indirectly checks whether the determination target subject is reflected in the corresponding video data 110c using the camera data 110e described in the section of the prior art. First, using the camera position, shooting direction, and angle of view of the camera data 110e, the shooting space of the camera is defined by an elliptical pyramid or a square pyramid. Next, it is determined whether or not the determination target subject exists in the photographing space using the position information of the existence region of the determination target subject in the subject region position information 110d. Specifically, by comparing the position of each point of the vector shape of the subject to be determined with the position of the imaging space, it is determined whether a part or all of the vector shape is included in the imaging space. If it is included in the shooting space, it is determined that the determination target subject exists in the shooting space, and the corresponding camera data 110e and the corresponding video data 110c are determined to be data showing the determination target subject.

  Note that, in the prior art, in practice, it is often determined whether or not the object exists within the imaging space by using the position (one point) of the determination subject instead of the shape of the determination subject region (a plurality of contour points). . In this embodiment, the subject region position information 110d of the present method is obtained from the vector map information 110a, and thus the shape (plural points) of the subject region to be determined is used. As shown in the principle diagram of the present invention in FIG. Since the subject area position information 110d does not matter whether it is one point or a plurality of points, it is possible to determine whether the subject area position information is inside the photographing space as one point as in the past.

  In the first embodiment, the camera data 110e is not passed as it is to the image determination method 110, but is passed through the conventional subject presence determination unit 60, and camera data whose determination target subject is not within the shooting range of each camera is not passed. The camera data in which the subject to be judged is not shown is not input to the image judgment method 110, and it is intended not to make a judgment on the image quality due to the shielding of other objects.

  The object shape estimation unit 120 includes the position information of the contour of the determination target object area in the object area position information 110d acquired from the vector map information 110a, and the building ground surface height map in the building / surface height map information 110b. The upper surface of the approximate shape of the subject to be determined is extracted by matching the information 110ba and examining the height of the ground surface of the building in relation to the position of each point of the contour. Similarly, the lower surface of the rough shape of the subject to be determined is checked by matching the surface height map information 110bb in the building / surface height map information 110b and examining the height of the ground surface related to the position of each point of the contour. Extract. In terms of height, an approximate shape is formed between the upper surface and the lower surface.

  FIG. 8 is an explanatory diagram for explaining processing in which the subject shape estimation unit 120 acquires the approximate shape of the determination target subject. As shown in the figure, the approximate shape of the subject to be judged is the corresponding vector figure (the figure indicated by a thick line. In the figure, the vector figure other than the subject to be judged is also indicated by a thin line for easy understanding. ) By adding the latitude / longitude value of an arbitrary point on the contour of), the height of the ground height map information 110ba of the building lump corresponding to the latitude / longitude value, and the height of the ground height map information 110bb, respectively. The approximate shape of the subject. The subject shape estimation unit 120 passes the extracted shape information of the determination target subject to the subject feature point extraction unit 130.

  Note that the method of extracting the approximate shape of the subject to be determined is not limited to the above-described method, and the height of each point on the contour of the subject is not accurately determined, for example, the center of the contour shape of the subject. The position may be obtained, and the height at the center position may be used as the height related to the subject.

  The subject feature point extraction unit 130 acquires the approximate shape information of the determination target subject from the subject shape estimation unit 120, and appropriately checks the height of the building and the height of the ground in the building / ground height map information 110b. Extract feature points of the subject to be determined. Hereinafter, the processing of the subject feature point extraction unit 130 will be specifically described. FIG. 9 is an explanatory diagram for explaining processing of the subject feature point extraction unit 130.

  As shown in the figure, the subject feature point extraction unit 130 first divides a side surface and an upper surface of the approximate shape of the determination target subject into meshes at arbitrary intervals. As the interval to be divided, one specified value (for example, 5 m) may be used, or a plurality of numerical values (for example, 1 m, 5 m, 10 m, etc.) may be prepared and divided by each numerical value.

  Next, the subject feature point extraction unit 130 determines the height of each lattice point on the upper surface of the divided mesh, in particular, the height of the building in the building / surface height map information 110b, that is, the height of the building indentation. The height is calculated while appropriately checking the height of the map information 110ba. Note that this height is not calculated accurately, as is the height of each point on the contour of the target object of the subject shape estimation unit 120, but the height of the representative position such as the center position of the contour shape is calculated. You may substitute as a common height of each lattice point.

  Furthermore, the subject feature point extraction unit 130 sets the height of each lattice point on the side surface of the divided meshes to the height of the upper surface of the building (= the height of the building incineration ground surface height map information 110ba) and the subordinate. The height of the ground (= the height of the ground surface height map information 110bb) is set to a value interpolated at a specified mesh interval.

  The subject feature point extraction unit 130 according to the first embodiment divides the lattice points of the mesh and the vertex groups of the approximate shape by dividing the side surfaces and the upper surface of the approximate shape of the determination target subject at arbitrary intervals in this way. A feature point group for the interval. The subject feature point extraction unit 130 passes the extracted feature point group information and subject shape information to the other object occlusion determination unit 140.

  Note that the method by which the subject feature point extraction unit 130 extracts feature points is not limited to the above-described method, and random values of height, latitude, and longitude are used as points included in the side surface or upper surface of the determination target subject. The acquired random values may be used as feature points.

  Further, the subject to be determined may be divided by strip division instead of mesh division, and each lattice point of the strip and the vertex group of the approximate shape may be used as the feature point group. Further, not only the side surface and the upper surface of the approximate shape but also points inside the approximate shape may be used as the feature points. Further, a single feature point may be extracted in order to reduce the amount of calculation related to the subject feature point extraction unit 130. In this case, for example, the point of the center of gravity (additional average position of all vertex position groups) of the determination target subject is used as the feature point. Further, it may be calculated so that the number of feature points in an arbitrary place, such as an upper layer part of a building that is more easily seen, is increased without obtaining the feature point group evenly with respect to the approximate shape.

  The other object occlusion determination unit 140 includes the shape information and feature point group information of the determination target subject from the subject feature point extraction unit 130, the building / surface height map information 110b, and the camera data 110e. Based on the three pieces of information of the camera data determined that the determination subject is reflected through the determination unit 60, it is determined whether there is a subject object existing between the camera and each feature point of the determination target subject. It is a processing unit that estimates whether or not the subject object is a determination target subject itself, thereby estimating hiding by another object.

  FIG. 10 is an explanatory diagram for explaining the processing of the other object occlusion determination unit 140. As shown in the figure, the other object occlusion determination unit 140 connects the camera position and the feature point with a line segment, and determines whether or not there is a subject that exceeds the connected line. Specifically, while following the line between the camera position and the feature point, the height of the building is examined using the height map information 110ba of the building litter of the building / surface height map information 110b. Go and look for a point where the height of the building is higher than the line segment. Next, whether or not the point is the determination target subject itself is determined from the shape information of the determination target subject, and if it is not the determination target subject itself, it is determined that it is another object.

  For example, as shown in FIG. 9A, the other object occlusion determination unit 140 finds a point M higher than the line segment between the feature point and the camera position on the ground height map information 110ba of the building litter. When the point M is not a point of the determination subject, it is determined that “another object exists between the camera and the feature point of the determination subject”. In such a case, a feature point that forms a line segment with the camera position is a feature point that is obstructed by another object (hereinafter referred to as a feature point (B)).

  Further, as shown in FIG. 10B, the other object occlusion determination unit 140 cannot find a point higher than the line segment between the feature point and the camera position on the surface height map information 110ba of the building litter. In this case, it is determined that “no other object exists between the feature points of the camera and the determination subject”. In such a case, a feature point that forms a line segment with the camera position is a feature point that is not obstructed by another object (hereinafter referred to as a feature point (A)).

  Further, as shown in FIG. 10C, the other object occlusion determination unit 140 finds a point N higher than the line segment between the feature point and the camera position on the ground surface height map information 110ba of the building litter. When the point N is a point in the subject, it is determined that “no other object exists between the camera and the feature point of the determination subject”. In such a case, a feature point that forms a line segment with the camera position is a feature point that is blocked by itself (hereinafter referred to as a feature point (C)).

  The other object occlusion determination unit 140 performs the above-described determination for each feature point group, counts the number of feature points (A), the number of feature points (B), and the number of feature points (C), Information relating to the number of points (A) to (C) is passed to the other object occlusion index calculation unit 150 as number information.

  The other object occlusion index calculation unit 150 acquires the number information from the other object occlusion determination unit 140, and calculates, as the other object occlusion index, the ratio that the determination target subject is obstructed by the other object based on the acquired number information. Is a processing unit. An example of a specific formula for calculating the ratio of obstruction by other objects is the ratio of obstruction by other objects = number of feature points (B) ÷ (number of all feature points (A + B + C) −feature points (C )). The other object occlusion index calculation unit 150 of the first embodiment uses the calculated ratio of occlusion by the other object as it is as the other object occlusion index and passes it to the display processing unit 160.

  Note that the ratio of obstruction by other objects and the method for calculating the final other object occlusion index are not limited to the above-described methods. For example, weighting is performed in consideration of differences in importance depending on feature points. You may make a calculation with In other words, in addition to the number information, the position information of each feature point is acquired from the other object occlusion determination unit 140, and in the case of a building that becomes a landmark, a feature point of a characteristic shape portion that becomes a landmark is obtained. In the case of an area where buildings are stood up, the hiding of the lower part of the building is not important. It is also possible to calculate.

  The display processing unit 160 acquires the other object occlusion index value of the determination target subject from the other object occlusion index calculation unit 150, and searches for video data 110c corresponding to the determination target camera data 110e based on the acquired index value. The processing unit reproduces and displays on the display device 50. FIG. 11 is a diagram illustrating an example of an interface in which the display processing unit 160 displays video and image data using an object occlusion index.

  As shown in FIG. 11, the display processing unit 160 according to the first embodiment of the present invention processes the video data (a, b, e) and the image data (c, d, f to k) in ascending order according to the values of the other object occlusion indices. (In order of decreasing numerical value to increasing numerical value. In this example, the smaller the numerical value, the lower the percentage obstructed by other objects). Furthermore, with the intention of recommending a subject that better reflects the judgment target subject, video and image data with other object occlusion index values less than a predetermined value are regarded as optimal video and image data for the judgment subject subject. For example, the corresponding images and thumbnail images of FIGS. 11A to 11E are displayed in color. On the other hand, video and image data whose other object occlusion index value is equal to or greater than a predetermined value are regarded as video and image data that are not appropriate for the subject to be determined. For example, the corresponding video and image data shown in FIGS. Display thumbnail images in black and white.

  In FIG. 11 of the first embodiment, the result is displayed in units of the original video and image data. By extracting only the good part, it may be divided into a plurality of video shots and displayed. In addition, the file names of videos and images to be recommended may be listed in a list, or highly recommended items may be displayed in a three-dimensional CG that is larger and closer to the screen, or an arbitrary display method may be used.

  The display processing unit 160 according to the first embodiment of the present invention browses and reproduces the selected video and image data when the user selects video data via the input device 40 with respect to FIG. And so on. FIG. 12 is a diagram showing an example of a display screen when the video data (a) is selected in FIG. In FIG. 12, among the image frame groups of the video, the conventional object presence determination unit 60 determines the image frame in which the determination subject exists in the shooting area of the camera, and the other object occlusion is performed by the image determination method 100 which is the present method. The portion of the image frame that is determined to have a low index (low shielding by other objects) is displayed in white on the black and white shading bar. By selecting only a white portion on the shading bar, the user can reproduce and view only a video frame that is recommended by the present method and that shows the determination subject without being blocked by other objects. Note that, as shown in FIG. 12, the reflected state of the determination subject obtained from the other object shielding index may be displayed with a light and dark bar, or the value of the other object shielding index may be displayed with a specific number. In addition, in the case of dividing the video shot into only the image frames with a low blocking rate at the stage of FIG. 11, the video shot to be reproduced in FIG. It is not necessary to display the index value and the image quality.

  Next, a processing procedure of the image determination apparatus 10 in the first embodiment will be described. FIG. 13 is a flowchart illustrating a processing procedure of the image determination apparatus 10 illustrated in the first embodiment. As shown in the figure, the image determination apparatus 10 aligns data used before and after the image determination method 100 (S102 to S106) and a portion that controls input and output at the same time. (S101, S104, S107).

  First, the subject region position information 110d is acquired from the designated vector shape information by designating a region related to the determination target subject in the vector map information 110a displayed on the display device 50 via the input device 40 (step S1). S101).

  Then, the subject shape estimation unit 120 specifies the approximate shape of the subject from the subject region position information 110d and the building / surface height map information 110b (step S102), and the subject feature point extraction unit 130 extracts the feature point group. (Step S103).

  Next, the conventional subject presence determination unit 60 uses the camera data 110e and the subject area position information 110d to sort the camera data in which the determination target subject is shown (step S104).

  Subsequently, the other object occlusion determination unit 140 executes the other object occlusion determination process at each feature point by using the determined camera data, the feature point group of the determined subject, and the building / surface height map information 110b. (Step S105).

  Subsequently, the other object occlusion index calculation unit 150 calculates an occlusion index value by the other object (step S106), and the display processing unit 160 outputs the video data 110c based on the occlusion index (step S107).

  Next, the other object occlusion determination process shown in step S105 of FIG. 13 will be described. FIG. 14 is a flowchart showing the other object occlusion determination process shown in step S105 of FIG. As shown in the figure, the other object occlusion determination unit 140 initializes the counts of feature points (A), (B), and (C) (the number of feature points (A) to (B) is set to 0). It is determined (step S201), and it is determined whether all feature points have been selected (step S202).

  When all the feature points have been selected (step S203, Yes), the number information is passed to the other object occlusion index calculation unit 150 (step S204). On the other hand, if not all feature points have been selected (No at step S203), unselected feature points are selected (step S205), and a line segment with the camera position as the start point and the selected feature point as the end point is selected. Create (step S206).

  Then, the other object occlusion determination unit 140 determines whether or not there is a subject exceeding the height of the created line segment (step S207), and when there is no subject exceeding the line segment (step S208, No) ), 1 is added to the count of the feature point (A) (step S209), and the process proceeds to step S202.

  On the other hand, if there is a subject exceeding the line segment (step S208, Yes), it is determined whether or not the subject exceeding the line segment height is the subject to be determined (step S210), and the line segment height is determined. If the subject exceeding the threshold is not the determination subject (step S211, No), 1 is added to the count of the feature point (B) (step S212), and the process proceeds to step S202.

  On the other hand, when the subject exceeding the height of the line segment is the subject to be determined (step S211, Yes), 1 is added to the count of the feature point (C) (step S213), and the process proceeds to step S202.

  As described above, the other object occlusion determination unit 140 creates the number information, the other object occlusion index calculation unit 150 calculates the object occlusion index using the number information, and the display processing unit 160 performs the other object occlusion index. Since the video data 110c is displayed on the display device 50 based on the above, the user can efficiently select the optimal video data.

  As described above, in the image determination device 10 according to the first embodiment, the subject shape estimation unit 120 uses the subject region position information 110d and the building / surface height map information 110b to specify the designated determination target subject. The shape is extracted, the subject feature point extraction unit 130 extracts the feature point based on the shape of the determination target subject, and the other object occlusion determination unit 140 detects the camera data 110e, the determination target subject shape, the feature point, and the building. Classifying number information of feature points related to other object occlusion is created based on the ground surface height map information 110b, and the other object occlusion index calculating unit 150 calculates another object occlusion index for each camera data based on the number information. To do. As shown in the first embodiment, the calculated other object occlusion index for each camera data is a reflection of the other object occlusion of the determination subject when the display processing unit 160 displays the related video data 110c on the display device 50. Since it can be used as the degree of recommendation for the condition, the optimum video data can be selected even when another object exists between the camera and the determination target subject.

  It should be noted that the building / surface height map information 110b according to the first embodiment includes the surface height map information 110bb which is the height of the pure ground, in addition to the surface height map information 110ba of the building litter. However, the ground surface height map information 110bb can be substituted only by the ground surface height map information 110ba of the building litter. In the first embodiment, the ground surface height map information 110bb is used by the subject shape estimation unit 120 to obtain the height of the lower surface of the subject shape, that is, the contact surface with the ground surface. Therefore, an example in which the subject shape estimation unit 120 acquires the lower surface of the subject, that is, the height information of the ground surface at the subject position, based on the subject region position information 110d and the ground height map information 110ba of the building dust. Show.

  The subject shape estimation unit 120 matches the building height map information 110ba with the subject region position information 110d, and does not include the subject height in the building height map information and is sufficiently larger than the subject region. In an area (there is rarely a point where there is no building somewhere in this sufficiently large area, since it is rare for buildings to touch each other without looking at the ground) Get the surface height of the corresponding building litter. Then, the subject shape estimation unit 120 sets the lowest height among the acquired heights as the height of the ground surface of the entire pseudo corresponding area. In addition, if the area where the height is acquired is too wide and the point considered as the height of the pseudo ground surface is far from the area where the subject exists, the error with the ground surface of the area where the subject exists increases. Instead of simply adopting the lowest height, the distance from the subject may be used for weighting.

  In this way, the subject shape estimation unit 120 acquires the height of the ground surface using the building dust height map information 110ba and the subject region position information 110d, so the ground surface height map information 110bb is obtained. There is no need to include it in the building / surface height map information 110, and it is possible to save the information recording capacity, the acquisition cost of the map information, and the time and effort of updating to keep the map information up-to-date.

  In addition, the object area position information 110d can be obtained by using the ground height map information 110ba of the building litter of the building / surface height map information 110b and substituting the vector map information 110a. Here, an example is shown in which the determination subject region position information 10d is acquired using the ground surface height map information 110ba of the building trash.

  FIG. 15 is an explanatory diagram for explaining a method of acquiring the subject area position information 110d from the ground height map information 110ba of the building litter. As shown in the figure, the subject to be judged is designated from the input device 40 or the like at one representative position, as in the conventional subject presence judgment. First, this representative position point is set as a representative point on the building litter surface height map information 110ba using a value such as latitude and longitude.

  And the height of the building litter ground surface corresponding to each point of the movement destination is acquired while moving from the representative point toward the periphery in a circular shape at predetermined intervals and angles. When obtaining the height of each point, the location where the height changes significantly compared to the height of the previous point (or the height of the representative point) (for example, the height of the representative point) And a place that has changed significantly is registered as an outer edge point of the determination subject region (in the example of FIG. 15, P1 and P2 are registered as outer edge points). When the outer edge points are searched for and registered toward the periphery while shifting the angle clockwise, the determined subject area of the subject area position information 110d is obtained by connecting the outer edge points in the registration order.

  In addition, although the example which acquires object area | region position information 110d was shown using only the building litter surface height map information 110ba among the building and ground surface height map information 110b here, surface height map information was shown. If the difference in height is also calculated using 110bb, the location where the height changes greatly, because the ground surface has undulations, the height of the ground of the building litter has changed even though the height of the building remains unchanged It is possible to more accurately estimate whether it is a place (such as a stepped apartment on an inclined land) or a place where the ground surface has no undulations and the height of the building itself has changed. In this case, in the former case, the outer edge point is not used, and only the latter case is used as the outer edge point, whereby a more accurate determination target subject region can be acquired.

  As described above, the object region position information 110d can be automatically acquired from the representative position of the determination subject using the building / surface height map information 110b without using the vector diagram information. The need to prepare and record the vector map information 110a can be eliminated, the recording capacity can be saved, and the subject region position information 110d can be automatically acquired by a simple designation method similar to that of the prior art. .

  Next, an apparatus (image determination apparatus) equipped with the image determination method according to the second embodiment will be described. The image determination apparatus according to the second embodiment performs clustering by hierarchically dividing the height data group of the building / surface height map 110b according to the first embodiment for each map area to which the image belongs, and the highest in the clustering result area. The minimum height is calculated. Then, the image determination apparatus efficiently performs other object occlusion determination using the highest height for each region, and calculates an other object occlusion index.

  FIG. 16 is a functional block diagram illustrating the configuration of the image determination device 20 according to the second embodiment. As shown in the figure, the image determination apparatus 20 includes a height clustering unit 210 and another object occlusion determination unit 220 having different internal operations. Since other configurations and operations are the same as those of the image determination device 20 shown in the first embodiment, the same components are denoted by the same reference numerals, and description thereof is omitted.

  The height clustering unit 210 specifies the maximum height and the minimum height for each predetermined area among the height group included on the map based on the building / surface height map information 110b (hereinafter, specified). The information is expressed as clustering information). FIG. 17 is an explanatory diagram for explaining the clustering information calculated by the height clustering unit 210.

  As shown in the figure, consider the case where the information of the building / surface height map information 110b is composed of height data for each minimum area unit, which is a height measurement unit. First, the height clustering unit 210 creates an integrated region in which four neighboring minimum region units are grouped together, compares the heights of the four minimum region units included in the region, and compares the heights in the integrated region. Specify the maximum height. In the example of FIG. 17, the area A21 is created by grouping the minimum area unit groups A211 to A214 having a height of 2 m to 4 m, and the maximum height is specified as 4 m. Similarly, A22 which summarizes the minimum area unit groups A221 to A224 having a height of 5 m to 6 m, A23 which summarizes the minimum area unit groups A231 to A234 having a height of 9 m to 12 m, and a minimum of 13 m to 16 m in height. Three regions A24, which are a group of region unit groups A241 to A244, are created, and the maximum height (6 m, 12 m, 16 m) is specified.

  Next, the height clustering unit 210 creates a new area A2 by collecting four neighboring areas A21 to A24, compares the maximum heights of the areas A21 to A24, and compares the maximum height of the area A2 ( 16m). In the example of FIG. 17, predetermined areas A1, A3, and A4 are created in the same manner and the maximum height is specified.

  The height clustering unit 210 repeatedly performs region integration and maximum height identification until the area of the integrated region that has been collected is equal to or greater than a specified area, or the integrated hierarchy by clustering is equal to or greater than the specified layer. The height clustering unit 210 passes the calculated clustering information to the other object occlusion determination unit 220.

  In the second embodiment, the height clustering unit 210 has shown an example in which four small areas in the vicinity are integrated to create one large area. However, the number of small areas to be integrated is not four. Alternatively, it may be created while dividing a large area. Furthermore, instead of integrating them uniformly, it may be created by changing the number of hierarchies or the number of integrated areas depending on whether it is an area with many search uses such as an urban area, or whether it is a place where the height difference changes rapidly. . Further, since the clustering process of the height clustering unit 210 can be performed regardless of the subject to be determined, the clustered data is held as the building / surface height map information 110b instead of performing the process every time. And can be used.

  The other object occlusion determination unit 220 creates number information related to the feature points (A), (B), and (C) based on the feature points extracted by the subject feature point extraction unit 130 and the clustering information. Part.

  FIG. 18 is an explanatory diagram for explaining a process of creating number information based on feature points and clustering information. FIG. 18 illustrates a case where the line segment T from the camera position to the determination target subject extends along the latitude. In addition, as shown in FIG. 18, a case will be described in which clustered regions A and B are included between the camera position and the determination target subject.

  First, the other object shielding determination unit 220 divides the line segment T into two along the areas A and B, and sets the divided line segments as TA and TB, respectively. Then, the other object shielding determination unit 220 compares the minimum heights TAMin and TBMin of the dividing line segments TA and TB with the maximum heights AMMax and BMax in the areas A and B.

  When the height of the line segment is higher than the heights of the regions A and B, that is, when TAMin> AMMax and TBMin> BMax, the other object occluding determination unit 220 has a feature point corresponding to the line segment. The point (A) is determined. This is because the line segment T is higher than the heights of all subjects existing in the areas A and B.

  On the other hand, when the height of the line segment is equal to or less than the height of the area A or B, that is, when TAMin ≦ AMMax or TBMin ≦ BMax (in the example of FIG. 17, TBMin ≦ BMax), the other object is shielded. The determination unit 220 compares the line segment T with the height of the subject included in the corresponding area by using the corresponding area in the area B (any one of B1 to B4).

  For example, when the line segment T corresponds to the areas B1 and B2 (when the line segment T passes over the areas B1 and B2, etc.), the subject height higher than the line segment T is used using the above-described method. It is determined whether or not there exists. Here, a case where B1 and B2 are units of the minimum area will be described. When a subject height higher than the line segment T does not exist in the areas B1 and B2, feature points corresponding to the line segment T are characterized. The point (A) is determined.

  On the other hand, when a subject height higher than the line segment T exists in the region B1 or B2, the feature point corresponding to the line segment T is determined as the feature point (B) or the feature point (C). When the corresponding area is a determination target subject, it is a feature point (C), and when the corresponding area is not a determination target subject, it is a feature point (B).

  The other object occlusion determination unit 220 classifies each feature point group related to the determination target subject into feature points (A), (B), and (C) by the above-described method, and the classified information is blocked as the number information. Pass to the rate calculator 150.

  In this manner, the other object occluding determination unit 220 compares the detailed height of each subordinate region unit and each measurement point with the height of the line segment only when necessary based on the clustering information. Therefore, it is possible to reduce the burden on the other object occlusion determination process.

  As described above, in the screen determination apparatus 20 according to the second embodiment, the height clustering unit 210 determines the height of the highest subject among the subjects based on the building / surface height map information 110b. The clustering information specified for each predetermined range is created, and the other object occlusion determination unit 220 creates the number information related to the feature point group using the clustering information, so that the optimum video data can be selected efficiently. it can.

  Note that when the integrated region is created by the height clustering unit 210, the minimum height of each integrated region may be calculated and used simultaneously with the maximum height. In the object shape estimation unit 120, the method described in the first embodiment of the present invention uses only the building litter surface height map information 110ba of the building / surface height map information 110b and substitutes the ground surface height map information 110bb. The technique for calculating the height of the lower surface of the subject, that is, the contact surface with the ground surface, can be more easily performed by using the minimum height of each integrated region by clustering. In other words, in the first embodiment of the present invention, in the area sufficiently larger than the determination subject area, the building-decomposed ground surface height map information 110ba is examined, the minimum height is calculated, and the ground surface of the entire pseudo relevant area is calculated. Although the technique considered to be the height was described, it is possible to efficiently calculate the minimum height of the corresponding area by obtaining the clustering integrated area group including the corresponding area from the position of the corresponding area and comparing the minimum heights thereof. it can.

  Next, an apparatus (image determination apparatus) equipped with the image determination method according to the third embodiment will be described. FIG. 19 is an explanatory diagram for explaining the concept of the image determination device according to the third embodiment. As shown on the left side of FIG. 19, subject A is photographed using camera A at time T1, subject B is photographed using camera B at time T2, and subject C is photographed using camera C at time T3. The image determination device determines that the image quality is good for any video data (the other object occlusion index is equal to or less than a predetermined value), but the video data is bad for the user. When the determination is made, the image determination apparatus determines that an obstruction has occurred in the overlapping portion of the imaging areas of the cameras A to C, and corrects the building / surface height map information and the like.

  Further, as shown on the right side of FIG. 19, the subject A is photographed at time T4, the subject B is photographed at time T5, and the subject C is photographed at time T6. Even if it is determined that the image quality is poor (the other object occlusion index is higher than a predetermined value), the image determination apparatus determines that each of the cameras A to It is determined that the shield that existed in the overlapping portion of the C imaging region has disappeared, and the building / surface height map information and the like are corrected.

  As described above, the image determination apparatus according to the third embodiment corrects the building / surface height map information and the like according to the response from the user related to the other object occlusion index, thereby improving the reliability related to the other object occlusion index. be able to. In addition, since it is not necessary to regularly update the building / surface height map information, the cost can be greatly reduced.

  FIG. 20 is a functional block diagram illustrating the configuration of the image determination device 30 according to the third embodiment. As shown in the figure, the image determination apparatus 30 includes a height map correction processing unit 330 and height map correction information 310a as related information of the building / surface height map information 110b. Since other configurations and operations are the same as those of the image determination apparatus shown in the first embodiment, the same components are denoted by the same reference numerals, and description thereof is omitted.

  The building / surface height map information 110b includes height map correction information 310a as related information. In addition, since the description regarding the building litter surface height map information 110ba and the ground surface height map information 110bb, which is an example of the building / surface height map information, is the same as that of the first embodiment, the description thereof is omitted. Here, in the third embodiment, the height map correction information 310a is part of the building / surface height map information 110b as an example. It doesn't matter.

  The height map correction information 310a is, among the building / surface height map information 110b, particularly height information related to the building, that is, correction information related to the surface height map information 110ba of the building trash here. . FIG. 21 is a diagram illustrating an example of the data structure of the height map correction information 310a. As shown in the figure, the height map correction information 310a includes a correction area table, a correction area shape table, and a correction area vertex table.

  The correction area table includes “ID”, “type”, “change occurrence estimated time (MIN)”, “change occurrence estimated time (MAX)”, “estimated shape ID”, “MBR (MIN) latitude / longitude”, “MBR”. (MAX) latitude / longitude ”and“ affiliation 3D mesh ID ”.

  Here, “ID (identification)” is information for identifying the correction data area, and “type” is an area (building) at a certain height due to an error that the height of the correction data area is too high. It is information on whether the occurrence of a certain area (building) was detected from the error that it was detected that it disappeared, or because it was too low, and “change occurrence estimated time (MIN)” and “change occurrence time estimated time” “(MAX)” is information indicating the time when the building has changed (disappeared or generated).

  The “estimated shape ID” is information for specifying the corresponding information in the correction area shape table, and “MBR (MIN) latitude / longitude” and “MBR (MAX) latitude / longitude” are the regions where the buildings exist. Information, “affiliation 3D mesh ID” is information for identifying the belonging 3D mesh code.

  Turning to the explanation of the correction area shape data, the correction area shape data includes “ID”, “estimated height (MAX)”, “estimated height (MIN)”, “number of projected shape vertices”, “projected shape vertex ID”. Group ". “ID” is identification information for associating with “estimated shape ID” related to the correction area table, and “estimated height (MAX)” and “estimated height (MIN)” are the estimated subject. It is height information.

  The “estimated height (MAX)” stores information on the maximum height in the area when it is determined that the building has disappeared, and the “estimated height (MIN)” When it is determined that a building has occurred, information on the minimum height in the area is stored. The “number of projected shape vertices” is information indicating the corresponding region shape or the number of vertices of the building included in the corresponding region, and the “projected shape vertex ID group” is information for identifying each vertex.

  The correction area vertex table has the latitude and longitude information of the vertices related to the estimated shape of the building, and includes “ID”, “latitude”, and “longitude”. Here, “ID” is information for associating with each ID of the “projection shape vertex ID group” related to the correction area shape table. “Latitude” and “longitude” respectively indicate latitude and longitude information.

  The subject shape estimation unit 120 calculates the height of the upper surface and the lower contact surface of the determination target subject region, and the subject feature point extraction unit 130 calculates the height of the feature point when extracting the feature point of the determination target subject. When the other object occlusion determination unit 140 searches for a point having a height higher than the line segment along the line segment connecting the camera and the feature point, the building of the building / surface height map information 110b is used. The height map correction information 310ba is used, and when the height map correction information 310a is collated at that time and the information of the predetermined area is corrected by the height map correction information 310a, the height map correction information 310a is used. Prioritize the information contained in.

  The height map correction processing unit 330 is a processing unit that receives a response to the other object occlusion index of the video data displayed by the display processing unit 160 and creates / updates the height map correction information 310a according to the received response.

  Specifically, when the height map correction processing unit 330 acquires from the input device 40 information (response) that the other object occlusion index of the video data displayed by the display processing unit 160 is incorrect, the height map correction processing unit 330 It is determined that the building / ground height map information 110b related to the area between the camera position where the video data was captured and the determination target subject has changed.

  Then, as described in the height map correction processing unit 330, for example, the left side of FIG. 19, when an area suspected of generating a shielding object can be specified to some extent, information regarding the area is included in the height map correction information 310a. Write. In this case, the “type” is set to “occurrence”, and the estimated range is registered in “MBR (MIN) longitude / latitude” and “MBR (MAX) longitude / latitude”.

  Further, regarding the time when the change is predicted, for example, in the example on the left side of FIG. 19, it can be estimated that the time has changed from time T1 to T3. Therefore, time T1 is registered in “change occurrence estimated time (MIN)”. T3 is registered in the “change occurrence estimated time (MAX)” (when T1 <T2 <T3).

  In addition, as described on the right side of FIG. 19, when an area suspected of disappearing a shielding object can be specified to some extent, information regarding the area is written in the height map correction information 310a. In this case, “type” is set to “annihilation”, and the estimated range is registered in “MBR (MIN) longitude / latitude” and “MBR (MAX) longitude / latitude”.

  Subsequently, regarding the time when the change is expected, for example, in the example on the right side of FIG. 19, it can be estimated that the time has changed from time T4 to T6. Therefore, the time T4 is registered in “change occurrence estimated time (MIN)” The time T6 is registered in the “change occurrence estimated time (MAX)” (when T4 <T5 <T6).

  As described above, in the image determination apparatus 30 according to the third embodiment, the height map correction processing unit 330 receives a response related to the other object occlusion index of the video data displayed by the display processing unit 160 from the input device 40. If the response includes information indicating that the other object occlusion index is incorrect, the height map correction information 310a is created / updated, so the building / surface height map information is updated periodically. Maintenance is eliminated and costs can be greatly reduced.

  In the example of the third embodiment, an example in which a shield is completely new and disappears is shown. However, in reality, a part of an invisible subject may be seen or hidden due to reconstruction of a building or the like. Conceivable. Also in this case, instead of superimposing at the shooting space level, we investigated in detail what part of each subject feature point group was determined to be invisible, etc. It is possible to estimate the change in the height of the area to some extent by obtaining the overlapping area and examining whether it should be higher or lower than the height of the line segment from which the area is calculated.

As described above, the image determination method according to the present invention captures a predetermined subject using a plurality of cameras and the like, and optimal video data is not burdened on the user from among the captured video data. This is useful for an image determination system that needs to be selected.

FIG. 1 is a principle diagram showing the principle of an image determination method according to the present invention. FIG. 2 is a diagram illustrating an example of the building / surface height map information. FIG. 3 is a functional block diagram illustrating the configuration of the image determination apparatus according to the first embodiment. FIG. 4 is a diagram illustrating an example of a vector map that is vector map information. FIG. 5 is a diagram illustrating an example of a screen on the display device that prompts the user to input using a vector map in order to acquire subject region position information. FIG. 6 is a diagram (1) illustrating another example of a screen on the display device that prompts the user to input using a vector map in order to acquire subject region position information. FIG. 7 is a diagram (2) illustrating another example of a screen on the display device that prompts the user to input using a vector map in order to acquire subject region position information. FIG. 8 is an explanatory diagram for explaining processing in which the subject shape estimation unit acquires the approximate shape of the determination target subject. FIG. 9 is an explanatory diagram for explaining processing of the subject feature point extraction unit. FIG. 10 is an explanatory diagram for explaining the processing of the other object occlusion determination unit. FIG. 11 is a diagram illustrating an example of an interface in which the display processing unit displays video and image data using an object occlusion index. FIG. 12 is a diagram showing an example of a display screen when the video data (a) is selected in FIG. FIG. 13 is a flowchart illustrating a processing procedure of the image determination apparatus illustrated in the first embodiment. FIG. 14 is a flowchart showing the other object occlusion determination process shown in step S105 of FIG. FIG. 15 is an explanatory diagram for explaining a method of acquiring subject area position information from the ground height map information of a building litter. FIG. 16 is a functional block diagram illustrating the configuration of the image determination apparatus according to the second embodiment. FIG. 17 is an explanatory diagram for explaining the clustering information calculated by the height clustering unit. FIG. 18 is an explanatory diagram for explaining a process of creating number information based on feature points and clustering information. FIG. 19 is an explanatory diagram for explaining the concept of the image determination device according to the third embodiment. FIG. 20 is a functional block diagram illustrating the configuration of the image determination apparatus according to the third embodiment. FIG. 21 is a diagram illustrating an example of the data structure of the height map correction information. FIG. 22 is an explanatory diagram for explaining the related art. FIG. 23 is an explanatory diagram for explaining the problems of the prior art.

Explanation of symbols

10, 20, 30 Image determination device 40 Input device 50 Display device 60 Conventional subject presence determination unit 100 Image determination method 110a Vector map information 110b Building / surface height map information 110ba Building litter surface height map information 110bb Ground surface height Map information 110c Video data 110d Determination subject region position information 110e Camera data 120, 320 Subject shape estimation unit 130 Subject feature point extraction unit 140, 220 Other object occlusion determination unit 150 Other object occlusion index calculation unit 160 Display processing unit 210 Height clustering Unit 310a height map correction information 330 height map correction processing unit

Claims (5)

A computer having a storage unit that stores a camera moving image obtained by photographing a subject and a storage unit that stores a line-of-sight position of the camera ,
A step of extracting a first feature point from among a plurality of feature points related to the shape of the subject based on information relating to the region where the subject exists;
Based on the first line segment that connects the position of the first feature point and the viewpoint position of the camera that captures the subject, and information on the existence area of the other subject, the other on the first line segment Determining whether or not a subject exists,
Determining whether the other subject exists on the second line segment based on a second line segment for a second feature point different from the first feature point;
Based on the determined result, the number of first line segments that overlap with other subjects among the plurality of first line segments, and the second line segment that overlaps with other subjects among the plurality of second line segments. By dividing the addition value obtained by adding the number by the number of the plurality of feature points, the ratio of the shooting of the subject by the camera being shielded by the other subject is calculated, and the display of the camera moving image is displayed. And a step of changing the ratio using the ratio . A step of calculating, for each different area, a maximum height among the heights of the other subjects in the area in the area included in the existing area of the other subject based on information relating to the subject existing region. Further, the computer executes , compares the maximum height for each region with the first line segment and the second line segment, and the first line segment and the second line below the maximum height for each region. Are determined as line segments that overlap the other subject, and the number of first line segments that overlap other subjects among the plurality of first line segments and a plurality of second segments are determined based on the determination result. A ratio is calculated by adding the number of second line segments that overlap with other subjects among the line segments and dividing by the number of the plurality of feature points, and the size of the ratio is determined by the predetermined subject. The image determination method according to claim 1, wherein the image is determined as a reflection state of the image.   Calculates the height of the lowest subject among multiple subjects as the height of the ground surface, corrects the height of the predetermined subject based on the calculated height of the ground surface, and then configures the shape related to the area where the predetermined subject exists The image determination method according to claim 1, wherein a feature point is extracted from the feature point group.   Scans map information that associates the position with the height of the subject or the ground surface, identifies the region of the subject based on the change in height, and shapes the predetermined subject based on the identified region and the height above the region The image determination method according to claim 1, wherein a feature point related to a predetermined subject is extracted from a feature point group constituting a shape. After determining that the subject is captured well based on the ratio of shooting of the subject by the camera being blocked by another subject , a response related to the determination result is received from the outside, and the determination is made based on the determination. The image determination method according to claim 4, further comprising: causing the computer to further execute the step of correcting the map information when information indicating that the result is incorrect is included.

Images